Digital Facial Morphology

3D Mimik-Archiv

A central goal of our research is the establishment of a morphological database of faces covering a broad range of inter-individual factors like sex, ethnicity but also variation related to expressions. Together with our cooperation partners from psychology, we are investigating new experimental paradigms to capture a broad range of facial performances, posed and spontaneous expressions for the collection of the 3D Mimik-Archiv.

Posed facial expressions (surface and photographic texture) from the 3D Mimik-Archiv captured in our Studio Camera Facialis.

Statistical Models of Facial Shape and Texture

Statistical shape models (SSMs) are well-known tools for the assessment and analysis of the morphology of human anatomy. Within this project, we are developing new algorithms that allow to automatically generate SSMs from large-scale databases for the statistical examination of inter- and intraindividual facial morphology.

To prevent a mixture of expression-characteristic motion patterns by a global PCA, we analyse the main modes of variation seperately per category. This results in an SSM with increased interpretability. Additionally, the preservation of characteristic motion patterns increases the accuracy of expression recognition.

Cultural Dimensions of Digital Facial Morphology

First work on the analysis of physiognomy is often dated back to the Greek philosophers Plato and Aristotle. Their work has been the foundation for the analysis of facial morphology and its critically discussed attribution to stereotypes and personal characteristics during history. Similarly, the analysis of affects and their manifestation in facial expressions has its roots in the work of the Greek philosophers. Through theories of affect and emotions, this association can still be identified in digital techniques for the coding of emotions from facial images today.

Together with our cooperation partners from the HU Cluster of Excellence and the Center for Literary and Cultural Research, we study these historical and cultural dimensions of current computer-aided methods. The aim is to identify the epistemic freight that is implicitly present in digital tools, such that it can be returned explicitly as part of our research.